Institutions and O¤shoring Decision

Marcella Nicolini

FEEM

FEEM Lunch SeminarFebruary, 4 2009

Motivation

I Huge literature suggests that institutions of the recipientcountry in�uence o¤shoring

I Contract theory suggests that institutional quality does nothave the same impact on di¤erent sectors

=) Does institutional environment a¤ect o¤shoring decisiondi¤erentially across sectors?

Motivation

I test whether institutional quality a¤ects production choicesabroad di¤erentially, depending on the sector

I show that institutions have a di¤erent impact according to theintended use of the shipped good (further manufacturing/resale)

I provide evidence using U.S. data on Direct Investment Abroad

Structure of the Presentation

1. Review of the Literature

2. Empirical Model

3. Data Description

4. Econometric Speci�cation

5. Results

6. Conclusions

Related Literature

Literature on FDI and institutional quality of the receiving country:

I Corruption: Wheeler Mody (1992) JIE , Hines (1995),Smarzynska Wei (2000), Wei (2000) REStat

I Democracy: Rodrik (1996), Harms Ursprung (2002), Busse(2004)

I Political stability: Juh Singh (1996), Stevens (2000), Jensen(2003), Busse Hefeker (2005)

I Financial institutions: Hermes Lensink (2003), Durham(2004), Alfaro Chanda Kalemli-Ozcan Sayek (2004) JIE

Related Literature

Contract theory suggests that institutional quality does not havethe same impact on di¤erent sectors:Institutions matter di¤erentially across sectors

I Empirical papers that refer to Rauch�s classi�cation (1999):Ranjan Lee (2005), Linders de Groot Rietveld (2005),Berkowitz Moenius Pistor (2006) REStat

I Institutional quality is a source of comparative advantage inthe production of complex goods: Costinot (2004), Levchenko(2007) RES, Nunn (2007) QJE, Acemoglu Antràs Helpman(2007) AER

Related Literature

Contracts shape the boundaries of the �rm:

I Hold-up problem shapes the boundaries of the �rm: McLaren(2000) AER, Grossman Helpman (2002) QJE, (2003) JEEA,(2005) RES, Ottaviano Turrini (2003), Ornelas Turner (2005)

I Hold-up problem also within �rm boundaries: Antràs (2003)QJE, Antràs Helpman (2004) JPE , Antràs Helpman (2006),Naghavi Ottaviano (2007)

I Empirical works on outsourcing: Bernard Jensen Schott(2005), Feenstra Hanson (2005) QJE, Swenson (2005) JIE,Nunn Tre�er (2007)

The Empirical Model

Institutional quality at country and industry level could a¤ect:

I outsourcing abroad (o¤shoring outside �rm boundaries)

=) Data are not available

I o¤shoring of production (hold-up also within �rm boundaries,see Grossman-Hart-Moore)

=) Data on multinationals

The Empirical Model

The o¤shoring decision is explained by:

I Standard determinants of FDI (vertical and/or horizontal)I Institutional quality at country and industry level

I estimate the following equation:

�owict = α+ instict + horict + verict + εict

�ow = intra�rm trade between parent company and foreigna¢ liate

i = 1...6 = sectorc = 1...62 = countryt = 1994, 1999 = year

The Data - Dependent Variable

Measures of o¤shoring:

Intra-�rm trade data from U.S. Direct Investment Abroad,Benchmark Surveys for 1994 and 1999, BEA

I have several measures:

- Total intra-�rm trade �ows between U.S. parent and foreigna¢ liates

It can be decomposed into:

- Trade �ows to U.S. parent company

- Trade �ows to foreign a¢ liates

I can further discriminate between:

- Goods shipped to foreign a¢ liates for further manufacturing

- Goods shipped to foreign a¢ liates for resale

The Data - Horizontal FDI

I control for determinants of horizontal FDI:

I market size: WDII tari¤s and NTBs: CEPIII freight and insurance costs: Feenstra World Trade FlowsI plant and corporate scale: U.S. Census Bureau

+ control for marginal corporate tax rate: WDI

The Data - Vertical FDI

I control for determinants of vertical FDI

Low cost of production are captured by comparative advantages(Yeaple 2003 REStat)

Interaction between:

I Country factor endowments: Hall Jones (1999)I Industry factor intensities: NBER productivity dataset

The Data - Institutional Quality

Institutional quality gives a comparative advantage in morecontract-dependent sectors

Institutions-driven comparative advantage is given by theinteraction between:

I Institutional quality at country levelI Institutional intensity at industry level

The Data - Institutional Quality

Institutional quality at country level:Kaufmann et al., World Bank:

I Voice and AccountabilityI Political Stability and ViolenceI Government E¤ectivenessI Regulatory BurdenI Rule of LawI Control of Corruption

The Data - Institutional Quality

Institutional quality at industry level:

I Nunn�s (2007) measure of contract intensity

instnri = ∑j θij

�Rneitherj + R ref . pricedj

�θij = value of input j in industry i / total value of all inputs usedin industry iRneitherj = proportion of input j not sold on an organized exchange,nor reference pricedR ref . pricedj = proportion of input j reference priced.Classi�cation of goods by Rauch (1999)

I Her�ndahl Index: measure of concentration in intermediateinput use (Levchenko, 2007)

The Choice of Estimation Technique

My dependent variable is integer and presents only positive values

Shapiro-Francia test rejects normality

=) I consider the family of Poisson Regression Models:

I PoissonI Zero-In�ated PoissonI Negative BinomialI Zero-In�ated Negative Binomial

The Choice of Estimation Technique

Descriptive statistics

Variable Nobs Mean Std. Dev. Min MaxTotal intrafirm trade 655 563.52 4239.36 0 81829Trade to U.S. parents 576 313.35 2378.82 0 44697Trade to foreign affiliates 589 320.23 2151.56 0 37132Further manifacturing 590 284.55 1870.35 0 35059Resale 630 17.76 322.88 0 7990

mean < variance=)Overdispersion

The Choice of Estimation Technique

0.1

.2.3

.4

0 2 4 6 8 10Count

Poisson ObservedZIP NBZINB

Observed and predicted distributions

The Choice of Estimation TechniqueComparison of alternative estimators

Dep Var: Total intrafirm flowsPoisson ZIP NB ZINB OLS

insti*instc 6.90*** 6.79*** 7.76*** 7.92*** 6.43***(.011) (.011) (.919) (.927) (.848)

constant 4.62*** 5.08*** 4.42*** 4.40*** 2.55***(.004) (.004) (.195) (.194) (.191)

Inflatedinsti*instc ­.347 128.32

(.595) (92.85)constant ­.526*** ­69.04

(.131) (49.78)

Log Likelihood ­846292 ­692923 ­2801 ­2798BIC 1688567 1381841 1590 1598AIC 2703 2213 8.96 8.96

Specification Tests 1690356 5.75 1.7e+06 0.700.000 0.000 0.000 0.242

Adjusted R2 0.12BP Test 7.95

(0.004)SF Test 6.21

(0.000)Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significantat 1%

The Results: Total intra �rm trade �owsDep. Var: Total intra­firm flows

(1) (2) (3) (4) (5) (6)insti*instc 7.093*** 4.552* 7.413*** 8.348*** 9.048*** 6.368**

(2.28) (2.35) (2.54) (2.69) (2.58) (3.08)instc 0.376 ­0.0393 0.0890 0.377 ­0.604 ­0.363

(0.74) (0.80) (0.91) (1.08) (1.08) (1.27)insti ­1.176 ­0.142 4.427** ­4.753** ­3.192 3.693

(1.44) (1.66) (1.82) (2.30) (2.29) (3.95)sales 1.722*** 1.832*** ­0.251 2.034*** 0.510 1.616**

(0.25) (0.48) (0.48) (0.32) (0.42) (0.67)market size 1.143*** 0.928*** 1.118*** 1.184*** 1.201*** 1.020***

(0.086) (0.13) (0.099) (0.087) (0.096) (0.13)tariff ­0.0114** ­0.0167***

(0.0049) (0.0057)corp. scale 0.0295* 0.0537*

(0.016) (0.031)plant scale ­0.0382*** ­0.0620**

(0.0088) (0.029)freight ­21.19*** ­23.16***

(3.69) (3.78)corp. tax rate 0.0136 0.0154

(0.020) (0.020)NTB 0.953 0.901

(0.61) (0.73)capitali*capitalc ­2.457 ­5.782*** ­8.592***

(1.54) (1.69) (2.11)capitalc 1.969* 4.363*** 6.277***

(1.10) (1.18) (1.46)capitali 38.33** 74.56*** 86.63***

(16.4) (17.9) (22.7)skilli*skillc ­53.83** ­82.08*** ­106.7***

(25.6) (29.1) (30.6)skillc 3.567 6.481** 8.742**

(2.81) (3.05) (3.42)skilli 57.74** 109.0*** 30.41

(23.2) (25.9) (47.3)constant ­47.03*** ­40.67*** ­52.70*** ­55.51*** ­97.58*** ­106.4***

(3.83) (6.69) (12.1) (5.11) (14.7) (17.9)LR test H0:α=0 1.4e+06 4.9e+05 1.3e+06 1.3e+06 1.3e+06 4.7e+05

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)Observations 603 425 585 585 585 425Log likelihood ­2641 ­2031 ­2613 ­2615 ­2597 ­2012Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%

The Results: Intra�rm trade to U.S. parentsDep. Var.: Intra­firm trade to U.S. parents

(1) (2) (3) (4) (5) (6)insti*instc 3.506 1.347 5.252 1.412 4.347 ­0.00853

(3.31) (3.39) (4.12) (4.05) (4.28) (4.48)instc 2.603** 0.919 2.038 4.518*** 3.571* 3.752*

(1.10) (1.19) (1.55) (1.72) (2.03) (2.14)insti 1.903 0.322 5.925** 0.857 0.603 14.92**

(2.08) (2.41) (2.59) (3.39) (3.35) (6.54)sales 1.665*** 3.119*** ­0.114 1.996*** 0.624 1.909**

(0.35) (0.82) (0.73) (0.46) (0.63) (0.97)market size 1.060*** 0.816*** 1.044*** 1.147*** 1.208*** 1.169***

(0.12) (0.21) (0.16) (0.13) (0.17) (0.20)tariff ­0.0214 ­0.0859***

(0.020) (0.027)corp. scale ­0.0233 ­0.00223

(0.028) (0.053)plant scale ­0.0129 ­0.0573

(0.016) (0.042)freight ­38.85*** ­45.14***

(8.39) (9.05)corp. tax rate 0.000650 0.0427

(0.040) (0.034)NTB ­2.017** ­0.938

(0.95) (1.05)capitali*capitalc ­1.615 ­5.353** ­13.31***

(2.38) (2.56) (3.38)capitalc 1.254 3.726** 8.558***

(1.74) (1.86) (2.40)capitali 28.45 68.79** 140.2***

(25.1) (27.3) (34.8)skilli*skillc ­31.61 ­63.77 ­185.9***

(34.0) (40.8) (45.9)skillc ­0.192 3.080 15.08***

(3.85) (4.44) (5.17)skilli 32.51 83.94** 27.47

(30.7) (36.4) (66.5)constant ­46.48*** ­52.24*** ­46.16** ­53.99*** ­91.40*** ­141.9***

(5.29) (11.3) (18.3) (7.02) (21.7) (28.0)LR test H0:α=0 7.3e+05 2.6e+05 7.1e+05 6.9e+05 6.7e+05 2.4e+05

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)Observations 526 365 509 509 509 365Log likelihood ­1683 ­1296 ­1672 ­1668 ­1662 ­1272Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%

The Results: Intra�rm Trade to Foreign A¢ liatesDep. Var.: Intra­firm trade to foreign affiliates

(1) (2) (3) (4) (5) (6)insti*instc 7.029*** 4.255** 6.224*** 9.105*** 8.680*** 6.319**

(2.11) (2.14) (2.24) (2.50) (2.31) (2.94)instc ­0.465 ­0.0639 ­0.480 ­0.960 ­1.746* ­1.382

(0.70) (0.74) (0.81) (0.95) (0.91) (1.19)insti ­2.112 ­0.474 4.413*** ­5.359** ­2.885 2.966

(1.34) (1.54) (1.67) (2.21) (2.19) (3.70)sales 1.635*** 1.065** ­0.374 1.858*** 0.260 0.958

(0.24) (0.43) (0.41) (0.30) (0.39) (0.64)market size 1.210*** 1.028*** 1.231*** 1.221*** 1.264*** 1.076***

(0.079) (0.11) (0.089) (0.083) (0.088) (0.12)tariff ­0.0108** ­0.0126**

(0.0048) (0.0058)corp. scale 0.0510*** 0.0532*

(0.014) (0.029)plant scale ­0.0360*** ­0.0510*

(0.0078) (0.027)freight ­18.88*** ­18.07***

(3.47) (3.60)corp. tax rate 0.00447 ­0.00993

(0.018) (0.019)NTB 1.775*** 1.321*

(0.60) (0.69)capitali*capitalc ­3.476** ­5.693*** ­6.109***

(1.42) (1.57) (1.81)capitalc 2.705*** 4.379*** 4.778***

(1.00) (1.09) (1.28)capitali 49.66*** 74.45*** 62.91***

(15.1) (16.6) (20.0)skilli*skillc ­43.59* ­69.21** ­74.70***

(25.3) (27.8) (28.9)skillc 3.632 6.431** 6.896**

(2.69) (2.83) (3.12)skilli 47.93** 99.19*** 20.59

(23.4) (25.0) (44.8)constant ­47.46*** ­34.90*** ­62.29*** ­53.84*** ­96.69*** ­85.36***

(3.64) (5.73) (11.1) (5.07) (13.8) (15.7)LR test H0:α=0 6.0e+05 2.3e+05 5.8e+05 5.7e+05 5.5e+05 2.2e+05

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)Observations 541 381 524 524 524 381Log likelihood ­2315 ­1788 ­2281 ­2295 ­2271 ­1778Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%

The Results: Goods for further manufacturingDep. Var.: Goods for further manufacturing

(1) (2) (3) (4) (5) (6)insti*instc 7.221*** 5.830** 6.533*** 9.224*** 8.892*** 8.708***

(2.15) (2.28) (2.27) (2.49) (2.29) (3.00)instc ­0.620 ­0.884 ­0.630 ­1.105 ­1.949** ­2.203*

(0.72) (0.80) (0.83) (0.95) (0.90) (1.23)insti ­2.147 ­1.167 3.939** ­5.982*** ­3.910* ­0.517

(1.37) (1.63) (1.69) (2.22) (2.19) (3.78)sales 1.627*** 0.943** ­0.257 1.912*** 0.432 0.867

(0.24) (0.43) (0.41) (0.30) (0.39) (0.64)market size 1.179*** 0.999*** 1.207*** 1.196*** 1.246*** 1.024***

(0.079) (0.12) (0.091) (0.083) (0.090) (0.12)tariff ­0.0123** ­0.0132**

(0.0050) (0.0061)corp. scale 0.0511*** 0.0531*

(0.014) (0.029)plant scale ­0.0353*** ­0.0420

(0.0078) (0.028)freight ­18.17*** ­17.04***

(3.49) (3.69)corp. tax rate 0.00870 0.00108

(0.018) (0.020)NTB 1.357** 0.925

(0.61) (0.69)capitali*capitalc ­3.459** ­5.784*** ­4.218**

(1.43) (1.57) (1.77)capitalc 2.658*** 4.425*** 3.365***

(1.00) (1.09) (1.25)capitali 48.74*** 74.91*** 43.54**

(15.2) (16.7) (20.3)skilli*skillc ­45.41* ­70.01** ­70.85**

(25.7) (28.3) (29.0)skillc 3.704 6.426** 6.299**

(2.73) (2.88) (3.11)skilli 53.78** 104.8*** 34.57

(23.7) (25.2) (46.3)constant ­46.57*** ­32.44*** ­62.03*** ­54.18*** ­98.66*** ­69.32***

(3.65) (5.68) (11.2) (5.08) (13.8) (15.7)LR test H0:α=0 5.3e+05 2.0e+05 5.2e+05 5.0e+05 4.9e+05 2.0e+05

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)Observations 542 385 526 526 526 385Log likelihood ­2275 ­1770 ­2245 ­2255 ­2233 ­1763Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%

The Results: Goods for resaleDep. Var.: Goods for resale

(1) (2) (3) (4) (5) (6)insti*instc 19.47*** 5.279 28.04*** 20.25 14.57 ­12.09

(6.90) (6.74) (9.05) (12.6) (11.5) (12.7)instc ­1.675 1.958 ­6.441 ­7.504* ­6.054 5.103

(2.82) (2.43) (4.21) (4.17) (4.24) (4.92)insti ­17.71*** ­6.250 ­9.005 ­11.56 4.107 12.47

(5.50) (5.37) (8.13) (10.1) (10.2) (14.0)sales 3.715*** ­1.148 ­0.758 2.137 ­1.876 ­1.447

(1.04) (1.43) (1.22) (1.50) (1.38) (2.20)market size 2.658*** 2.576*** 2.398*** 2.578*** 2.490*** 2.913***

(0.50) (0.51) (0.44) (0.46) (0.44) (0.59)tariff ­0.0183 ­0.0431

(0.042) (0.047)corp. scale 0.183*** 0.199*

(0.037) (0.10)plant scale ­0.0344 ­0.0630

(0.030) (0.096)freight ­10.50 ­8.581

(15.3) (15.7)corp. tax rate 0.0669 0.0521

(0.055) (0.062)NTB 0.666 0.211

(2.10) (2.18)capitali*capitalc 3.505 5.378 4.230

(5.15) (5.92) (5.82)capitalc ­1.485 ­3.536 ­3.719

(3.91) (4.54) (4.47)capitali ­8.325 ­22.65 ­48.76

(56.4) (63.8) (66.1)skilli*skillc ­66.78 121.7 210.4**

(117) (117) (101)skillc 13.09 ­4.708 ­16.10

(11.1) (10.9) (10.00)skilli ­1.395 ­117.9 ­243.2

(105) (98.2) (164)constant ­113.3*** ­59.77*** ­56.65 ­95.21*** ­23.50 ­5.554

(17.5) (20.5) (46.8) (23.6) (55.4) (50.0)LR test H0:α=0 7.2e+04 9380.19 6.6e+04 3.5e+04 1.6e+04 5427.01

(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)Observations 577 396 559 559 559 396Log likelihood ­299.1 ­254.2 ­290.5 ­296.0 ­287.4 ­251.1Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%

The Results: Marginal e¤ects for the institutional term

(1) (2) (3) (4) (5) (6)Trade to foreign af. 466.27*** 335.69* 384.69*** 642.14*** 489.71*** 444.05**

(147.02) (173.27) (143.34) (186.97) (138.58) (211.55)Further manif. 446.43*** 419.55** 376.8*** 598.36*** 459.87*** 571.67***

(139.86) (171.04) (136.6) (171.77) (127.01) (208.69)Resale 1.688 0.450 1.533 1.960 0.623 ­0.733

(1.085) (0.623) (1.075) (1.636) (0.713) 0.688Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%

Robustness

I consider panel estimates in order to exploit the time-varyingdimension of the dataset

country or sector e¤ects

estimator: Panel Negative Binomial

RobustnessPanel Estimates: Fixed vs Random E¤ects

insti*instc 3.775*** 3.777*** 3.706*** 3.739*** 3.571 2.963(0.99) (0.99) (0.99) (0.99) (3.72) (3.64)

insti ­3.332*** ­3.299*** ­3.216*** ­3.210*** ­5.830** ­5.348*(0.71) (0.71) (0.71) (0.71) (2.85) (2.75)

instc ­0.739** ­0.670* ­0.855** ­0.786** 0.852 ­0.252(0.36) (0.35) (0.37) (0.36) (1.49) (1.09)

sales 0.349*** 0.349*** 0.373*** 0.371*** 0.891** 0.874**(0.10) (0.10) (0.10) (0.10) (0.35) (0.35)

market size 0.281*** 0.323*** 0.300*** 0.338*** 0.320 0.785***(0.053) (0.051) (0.053) (0.050) (0.31) (0.16)

constant ­11.87*** ­13.02*** ­12.64*** ­13.68*** ­22.33** ­33.90***(1.88) (1.82) (1.86) (1.81) (9.09) (5.97)

Log likelihood ­1790 ­2205 ­1761 ­2171 ­160.3 ­288.2Estimation FE RE FE RE FE REStandard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%;Estimates with country effects

Intrafirm trade to foreignaffiliates

Goods for furthermanifacturing

Goods for resale

Robustness

Panel Estimates with Country Fixed E¤ects

Total intra­firmflows

Intra­firm trade toU.S. parents

Inta­firm trade toforeign affiliates

Goods for furthermanufacturing Goods for resale

insti*instc 4.101*** 2.108 3.775*** 3.706*** 3.571(0.98) (1.42) (0.99) (0.99) (3.72)

insti ­3.276*** ­1.096 ­3.332*** ­3.216*** ­5.830**(0.71) (1.04) (0.71) (0.71) (2.85)

instc ­0.761** 0.500 ­0.739** ­0.855** 0.852(0.35) (0.53) (0.36) (0.37) (1.49)

sales 0.337*** 0.335*** 0.349*** 0.373*** 0.891**(0.100) (0.13) (0.10) (0.10) (0.35)

market size 0.288*** 0.371*** 0.281*** 0.300*** 0.320(0.051) (0.062) (0.053) (0.053) (0.31)

constant ­12.14*** ­15.75*** ­11.87*** ­12.64*** ­22.33**(1.79) (2.27) (1.88) (1.86) (9.09)

Observations 586 395 525 527 230Log likelihood ­2067 ­1209 ­1791 ­1761 ­160.3Panel estimates with country fixed effects. Standard errors in parentheses; * significant at 10%; ** significantat 5%; *** significant at 1%

Robustness

Panel Estimates with Sector Fixed E¤ects

Total intra­firmflows

Intra­firm tradeto U.S. parents

Inta­firm trade toforeign affiliates

Goods for furthermanufacturing

Goods for resale

insti*instc 3.032*** 1.686 2.778** 2.929*** 1.267(1.08) (1.51) (1.11) (1.10) (4.16)

insti ­4.002*** ­2.154* ­4.111*** ­4.210*** ­5.963*(0.78) (1.10) (0.80) (0.80) (3.17)

instc ­0.254 0.816 ­0.224 ­0.403 0.741(0.34) (0.52) (0.34) (0.35) (1.10)

sales ­0.0303 0.0392 ­0.0834 ­0.0425 ­0.442(0.12) (0.15) (0.12) (0.12) (0.46)

market size 0.500*** 0.547*** 0.525*** 0.527*** 0.766***(0.037) (0.045) (0.037) (0.037) (0.11)

constant ­13.61*** ­17.28*** ­13.37*** ­13.80*** ­17.45***(1.67) (2.13) (1.74) (1.75) (6.51)

Observations 603 526 541 542 577Log likelihood ­2538 ­1569 ­2217 ­2176 ­243.7Panel estimates with sector fixed effects. Standard errors in parentheses; * significant at 10%; ** significantat 5%; *** significant at 1%

Sensitivity

Results with Her�ndahl Index

Total intra­firmflows

Intra­firm trade toU.S. parents

Inta­firm trade toforeign affiliates

Goods for furthermanufacturing Goods for resale

insti*instc 1.990* 0.597 2.282** 2.155** ­0.867(1.03) (1.47) (1.03) (1.04) (6.23)

insti ­0.434 0.951 ­0.842 ­0.789 2.497(0.73) (1.06) (0.70) (0.71) (4.98)

instc 1.531** 3.414*** 0.571 0.562 3.777(0.65) (0.98) (0.62) (0.62) (3.30)

sales 2.320*** 2.551*** 2.059*** 2.047*** 3.200***(0.24) (0.34) (0.22) (0.22) (1.06)

market size 1.139*** 1.161*** 1.197*** 1.158*** 2.426***(0.083) (0.12) (0.078) (0.078) (0.43)

constant ­54.47*** ­59.93*** ­52.61*** ­51.52*** ­106.6***(3.79) (5.45) (3.53) (3.57) (15.5)

LR test H0:α=0 1.4e+06 7.5e+05 6.2e+05 5.4e+05 7.0e+04(0.000) (0.000) (0.000) (0.000) (0.000)

Observations 603 526 541 542 577Log likelihood ­2650 ­1688 ­2321 ­2282 ­303.1Standard errors in parentheses; * significant at 10%; ** significant at 5%; *** significant at 1%

Conclusions

I Institutions do matter in o¤shoring decisionI Impact is di¤erent across sectors: institutions are moreimportant in the production of more complex goods

I Empirical evidence using di¤erent measures of intra �rm tradeI Institutional quality does not impact when considering �nalgoods

Conclusions

Results are robust using:

I di¤erent measures of intra-�rm tradeI di¤erent control variablesI di¤erent speci�cations (pooled/panel)I di¤erent measures of institutional quality at country andindustry level